Rapid advancements in Artificial Intelligence (AI) continue to introduce the technology to a wide range of industrial applications – with the medical device industry being no exception. Advances in analytical and computing techniques, coupled with the explosion of data in healthcare organizations, bring about many potential use cases of artificial intelligence to the fields of medicine and healthcare. As the presence of AI becomes increasingly adopted we can start to analyze the benefits of artificial intelligence in medicine and healthcare.
Examples of some of the advantages to consider, include uncovering leading clinical practices, reducing research discovery time, streamlining administration and offering new personalized patient treatment. Embraced at an industrial scale, AI has the potential to align patients and caregiver’s decisions in ways that improve treatment outcomes and add value.
Additionally, investors are also realising the benefits of healthcare AI and continue to invest to fund advancement in the field.
As reported by Fierce Healthcare, in 2019, AI-based startups raised $26.6B across 2,235 investment deals, with healthcare AI startups being the most funded industry of all.
In this article, we are going to investigate the potential benefits of artificial intelligence in healthcare and some of the already existing and growing applications of AI in the healthcare and medical fields.
Existing examples of AI in action
There are already many examples of AI appearing within everyday life applications. A good example of this is the Nest devices for home automation, using AI to determine personal preferences and adjustments for our everyday needs.
Google Assistant on your smartphone is another example, supplying users with helpful “just in time” information based on what it learns about your preferences and past user behaviour.
Self-driving cars – popularized by Tesla – are another example of AI in action.
All of these examples use different aspects of AI to enable technology to be integrated in different applications. It is important to understand the difference between these terms and how they apply to different applications.
Machine Learning and Artificial Intelligence
AI is a broad term used to describe any human-like intelligence displayed by computers, but it doesn’t necessarily differentiate between specific terminology, such as machine learning, deep learning or natural language processing (NLP).
As demonstrated in the below image, artificial intelligence can be broken down into main and sub-branches of study.
One aspect of artificial intelligence that shows great promise within healthcare and medical fields is machine learning.
Machine learning is the application of AI built using smart algorithms. As more data is inputted into the system, the machine learns from it without relying on rule-base programming.
For example, clinicians could use known data – such as medical images – that have been assigned a ‘disease’ or ‘healthy’ state. Inputting known data into machine learning algorithms helps the machine then ‘learn’ the differences between the two. Future images inputted into the algorithm would then be able to determine one of the two states based on existing data submitted to it.
Deep learning takes machine learning one step further, creating structured algorithm layers to create what is known as an artificial neural network. The output from deep learning systems allow for greater accuracy at influencing decision making, in a system that can train itself and with less human intervention, as referenced above.
AI has already proven itself to be quite the disruptor – bringing about a whole new paradigm for how things work or how we get tasks done. Let’s take a look at the benefits of AI and how it can be adopted and applied to a range of healthcare and medical device applications.
Improving Clinical Health Data Management
As health care moves toward adopting digital health, the requirement for generating and collecting more data is required. Clinicians and Healthcare Professionals already struggle with data overload from a myriad of healthcare and connected medical devices. Providing them with more raw data will only serve to overwhelm and prove ineffective to advancing care.
One way to overcome the data overload associated with digital health applications is the implementation of smart algorithms. These machine learning algorithms can provide insights by processing the data and identify patterns that are not immediately obvious to the human eye; a more efficient way of getting actionable data without relying on human intervention.
A good example of smart algorithms in action are devices that monitor glucose levels. Rather than providing the raw data of glucose levels, these devices identify trends in and only notify the user when their action is required. Advanced versions of these devices – such as Medtronics’ Continuous Glucose Monitoring device (CGM) – are capable of automatically adjusting insulin doses to respond to key glucose information.
This is just one of many examples where companies are developing new digital health technology with similar principles for other monitored conditions.
Processing large data sets for diagnosis
The processing of large amounts of data is another benefit of AI in healthcare and an area where AI is disrupting the medical device world.
For example, a recent NIH study into the diagnosis of metastatic breast cancers using artificial intelligence found that the technology was as accurate as 99%. In this study, the AI was even proven to be able to detect the presence of spreading cancer cells – micrometastases – which are typically very difficult for human pathologists to detect.
Another study found that, when diagnosing on their own (without the aid of AI), pathologists miss up to 60% of small tumors. This is another way in which AI can prove to be a lifesaver when used with medical devices. Early detection and appropriate treatment are always preferable for better health outcomes.
These examples are not a case of using AI technology to replace healthcare professionals, but an example where using new technology can provide better diagnosis and care, at a greater speed and with less human intervention.
Improving healthcare in under-resourced areas
Some countries face particular challenges from being under-resourced with qualified clinicians and/or have less access to skilled training. In these countries, undiscovered or misdiagnosis of health issues tend to be more common.
In one study, the use of high-resolution microendoscope images for the diagnosis of esophageal tumors was found to be highly effective, with great potential for use in countries with skill or resource challenges.
In contrast, NASA’s Human Research Program is developing a platform that uses machine learning to identify a wide variety of issues that are seen as critical issues for space flight.
This new technology allows important factors such as bone density, intracranial pressure, and cardiovascular pathologies to be continually monitored in a hostile environment such as space.
NASA is also developing smart guidance systems that allow relatively untrained astronauts to use ultrasound machines properly. By providing “GPS-like” guidance, astronauts know where to move the probe and how to move it to get good images.
These same technologies developed by NASA could again be applied to any under-resourced communities. If the technology is proven accurate, then this could help speed accurate diagnosis for these populations.
Speeding up Drug Development with Artificial Intelligence
The traditional development process takes a very long time before a drug is successfully developed, often involving several “misses” before companies strike one formula that works. This long development period adds to the high cost of drugs that we see today.
By implementing AI within the drug development process, scientists hope to identify the most promising developments early on within the R&D process, helping to save time and resources as part of the initial discovery phase.
One example of AI being used within the drug development process is demonstrated in a partnership between GNS Healthcare and REFS (Reverse Engineering and Forward Simulation) to unlock and analyze complex medical data. Using patient data, researchers can generate new models to find hidden drivers in cancer progression.
Another example of AI in action, is the partnership between GSK and Exscientia to discover novel and selective small molecules for up to 10 disease-related targets. The aim is also to make the early stages of drug discovery more efficient.
In fact, if we look at the entire development cycle for pharmaceuticals, AI can have an impact at every phase.
For patients who have exhausted all other treatments for their condition, a clinical trial may be their last hope. One of the major challenges for patients is finding and being accepted onto a relevant clinical trial.
Currently, patients have to scour the government database of clinical trials themselves, unless their physician or someone they know happens to already know of a trial for them. They then go through an exhaustive process of evaluations for inclusion and exclusion criteria, as detailed in the process map below from CB Insights.
The many benefits of AI in healthcare doesn’t stop at physicians, but can also be applied to patient impact.
Many patients miss out due to enrollment difficulties, so AI can help by recommending trials that are a match for patient conditions or symptoms. By using smart algorithms to extract data from a patient’s medical records, suitable study recommendations can be made, saving the patient hours of painstaking searching for a trial.
AI in the medical field & medical device design
Another application of AI is to monitor and assess information that is pertinent to a medical device’s design, in the hopes of creating improved designs in the future.
Using AI, the improved ability to monitor issues such as adverse events, complaints and related conditions, is beneficial for future product designs. With machine learning, using algorithms to scan product databases assist with future product designs that can address concerns as part of the product development process.
Current challenges for artificial intelligence within healthcare
For artificial intelligence to be successful with healthcare, it is vital that data management is a top priority.
In order for companies to stay compliant with laws and regulations such as HIPAA or GDPR, the correct access, storage and security of data is paramount for any systems to be successfully implemented.
Overcoming compliance challenges can be complicated further when there are additional factors to solve, such as existing software and infrastructure that is used to manage data from a range of different platforms and networks.
The complexity of artificial intelligence in the healthcare and medical industries is greater than outlined in this article.
Fortunately, there are plenty of additional resources to help stay up-to-date and understand the complexities, solutions and news within the field:
- Galen Data – Cloud Solutions for Medical Device Connectivity
- FDA – Software as a Medical Device (SaMD)
- FDA – Proposed Regulatory Framework for Modifications to Artificial Intelligence/
- FDA – Evolving Role of Artificial Intelligence in Radiological Imaging
- Machine Learning (AI/ML)- Based Software as a Medical Device (SaMD)
- IBM Watson – Artificial Intelligence in Medicine
- Health IT Analytics
- The Brookings Institute – Risks and Remedies for Artificial Intelligence in Health Care
- Johner Institute – AI in Medicine
- Gartner – The Need for AI Governance in Healthcare
Summarizing AI in Healthcare and Medical Fields
AI is one of the biggest disruptors of our generation and has the potential to be a huge asset to medical device companies in future years. Like with most disruptive technologies, there are many potential pitfalls that could prevent the successful implementation of AI within the industry.
The number of benefits for artificial intelligence in healthcare is vast; whether that be reducing the reliance of human knowledge, speeding up the drug development process, or making healthcare cheaper and accessible to a range of different populations.
For artificial intelligence to be a success in any industry, smart algorithms need unfiltered access to a range of data sets. This is a challenge for the healthcare and medical device industry, as regulations such as HIPAA make it challenging to share data whilst remaining compliant with laws and regulations.
Despite these sorts of challenges, AI continues to progress as healthcare companies focus on expanding their digital capabilities.
As clinicians, healthcare professionals and patients start to become aware of the benefits associated with AI, innovation in the field will continue, making artificial intelligence in healthcare cheaper and more accessible to patients worldwide.